Re: Concurrency Control in Clustered Web Application. Any pattern some one ?

From: Lennart Jonsson <lennart_at_kommunicera.umea.se>
Date: 18 Jan 2004 02:18:19 -0800
Message-ID: <6dae7e65.0401180218.19f82b71_at_posting.google.com>


aerosteak_at_hotmail.com (Hugo Villeneuve) wrote in message news:<a93284e4.0401171722.7c2ef307_at_posting.google.com>...

[...]

> But then, what happens if a server goes down ? All locked requests
> will stay locked forever. Then we thought that we could unlock all
> requests when we start the Web Server. It will work well if we had
> only one server but we have more then one.
>

One idea is to timestamp the locks and have the client refresh it's locks every now and then.  

> Looks like we are not on the right track. Is there a simpler way to
> manage Concurrency Control ? Any advice will be welcome.

Here is one article describing an opposite aproach. Perhaps you can make something out of that:

http://www-106.ibm.com/developerworks/db2/library/techarticle/0209kuznetsov/0209kuznetsov.html Received on Sun Jan 18 2004 - 11:18:19 CET

Original text of this message